TGM2 A Cell Surface Marker in Esophageal Adenocarcinomas

Abstract

Introduction

Esophageal adenocarcinomas (EAC) are aggressive cancers that are increasing in incidence and associated with a poor prognosis. The identification of highly expressed genes in EAC relative to metaplastic Barrett’s esophagus (BE) may provide new targets for novel early cancer detection strategies using endoscopically administered, fluorescently labeled peptides.

Methods

Gene expression analysis of BE and EACs were used to identify the cell surface marker transglutaminase 2 (TGM2) as overexpressed in cancer. The expression of two major isoforms of TGM2 was determined by qRT-polymerase chain reaction in an independent cohort of 128 EACs. Protein expression was confirmed by tissue microarrays and immunoblot analysis of EAC cell lines. TGM2 DNA copy number was assessed using single nucleotide polymorphism microarrays and confirmed by qPCR. TGM2 expression in neoadjuvantly treated EACs and following small interfering RNA-mediated knockdown in cisplatin-treated EAC cells was used to determine its possible role in chemoresistance.

Results

TGM2 is overexpressed in 15 EACs relative to 26 BE samples. Overexpression of both TGM2 isoforms was confirmed in 128 EACs and associated with higher tumor stage, poor differentiation, and increased inflammatory and desmoplastic response. Tissue microarrays and immunohistochemistry confirmed elevated TGM2 protein expression in EAC. Single nucleotide polymorphism and qPCR analysis revealed increased TGM2 gene copy number as one mechanism underlying elevated TGM2 expression. TGM2 was highly expressed in resistant EAC after patient treatment with neoadjuvant chemotherapy/radiation suggesting a role for TGM2 in chemoresistance.

Conclusion

TGM2 may be a useful cell surface biomarker for early detection of EAC.

The incidence of esophageal adenocarcinoma (EAC) has increased rapidly in Western countries at a rate greater than most common cancers.¹ This increase is thought to reflect a rise in obesity, chronic gastroesophageal reflux disease, and the development of the metaplastic condition termed Barrett’s esophagus (BE), where the normal esophageal squamous epithelium is replaced by an intestinal-type columnar epithelium.² Carcinomas arising at the gastroesophageal junction (GEJ) are also increasing although these do not seem to develop from Barrett’s metaplasia. Unfortunately, most patients are diagnosed with EAC or GEJ cancers at more advanced stages and when current treatments including chemotherapy and radiation are less effective. When patients are diagnosed with either early stage cancer or the high-risk BE called high-grade dysplasia, surgical or endoscopically based interventions are associated with much more favorable patient outcomes.³ Current surveillance approaches for BE have limitations.⁴ As recently highlighted in a review by Vogelstein and colleagues,⁵ significant improvement in survival rates for cancer patients require the development of more effective approaches for early cancer detection. As a major advance in this direction, we have incorporated detection tools for EAC based on fluorescently labeled peptides directed against cell surface proteins highly expressed in EAC allowing for imaging and early cancer detection with innovative noninvasive optical endoscopy.⁶

We have utilized expression profiling to examine genes that are overexpressed in EAC relative to Barrett’s metaplasia or dysplasia.^7,8 Genes that are cell membrane targets or associated with the cell surface could provide important candidates for peptide-directed early cancer detection.⁹ Other strategies have incorporated fluorescent lectin molecules for the identification of dysplastic BE.¹⁰ As identified in the present study, tissue transglutaminase 2 (TGM2) is a highly expressed cell surface protein in EAC and a multifunctional molecule that can hydrolyze GTP and catalyze the covalent crosslinking of extracellular proteins.¹¹ TGM2 is the most abundantly expressed member of the transglutaminase family known to have two common isoforms and multiple alternatively spliced minor forms.¹² Little is known regarding TGM2 isoform function or their regulation; however, alternative splicing seems to be an active process in cancer.¹³

TGM2 was recently reported to be overexpressed in esophageal squamous cell carcinomas, and it was suggested that it may function as a tumor growth suppressor when bound to the G-protein receptor GPR56,¹⁴ and a reduced interaction of the two is hypothesized to release cells from a quiescent state to a more metastatic state.¹⁵ In other epithelial cancers cell lines, expression of TGM2 has been linked to cisplatin and doxorubicin drug resistance with knockdown of TGM2 causing increased chemosensitivity.^16–18 TGM2 is also overexpressed in various primary tumors including lung,¹⁹ colon,²⁰ ovarian,²¹ breast,⁶ and glioblastoma.²² The role of TGM2 and its various isoforms in these cancers may differ as to whether it induces apoptosis,²³ cell proliferation,²⁴ or chemoresistance,^25–27 and the exact mechanisms of TGM2 regulation is an area of intense investigation.²⁸

In the present study, we have identified TGM2 as a possible cell surface marker for EAC that may be useful for its development as a diagnostic marker. We measured the expression of the major isoforms of TGM2, validated its mRNA expression in an independent cohort of EACs and also examined TGM2 protein overexpression using tumor tissue microarrays (TMAs). We determined that the overexpression of TGM2 in EAC is in part because of gene amplification at 20q. In the neoadjuvant setting, EAC patients treated with chemotherapy and radiation do significantly worse when their cancers are still present at the time of surgery. We observe that these “resistant” EACs show increased expression of TGM2 at a frequency higher than those of untreated EACs, suggesting elevated TGM2 may be associated with reduced responsiveness to standard neoadjuvant therapy and that TGM2 overexpression in EAC therefore may be useful as a potential marker for early cancer detection or an indicator for chemotherapeutic resistance.

MATERIALS AND METHODS

Cell Lines and Reagents

OE19 and OE33 EAC cell lines were obtained from Sigma-Aldrich and cultured in RPMI media (Invitrogen, Carlsbad CA) supplemented with 10% fetal bovine serum (Atlanta Biologicals, Norcross, GA) and 1% antibiotic-antimycotic (GIBCO) at 37°C in a 5% CO₂ atmosphere. Methyltransferase inhibitor 5′-aza-2′-deoxycytidine (A3656, Sigma, St. Louis, MO) and histone deacetylation inhibitor trichostatin A (T8552, Sigma) were used as a 5 mM stock in DMSO and stored at −20°C. Cisplatin (P4394, Sigma) made as a 5 mM stock in phosphate-buffered saline was used immediately. Transforming growth factor β (100–21, PeproTech, Rocky Hill, NJ) stock was dissolved in 10 mM citric acid pH 3.0, at a 10 ng/μL stock and stored at −80°C. Cell lines were genotyped for authenticity using the Identifiler Plus kit (Applied Biosystems, Grand Island NY) at the University of Michigan DNA sequencing core facility (P01 HL057346).

Patients and Samples

Patient written consent was obtained and the study received approval from the University of Michigan Medical School Institutional Review Board. One-hundred and twenty-eight patients used in this study did not receive preoperative radiation or chemotherapy. Tissues were obtained from patients undergoing esophagectomy for cancer or highgrade dysplasia (HGD) at the University of Michigan Health System. Twenty-one patients who had been treated with neoadjuvant chemotherapy (cisplatin and 5-fluorouracil or carboplatin and paclitaxel) and radiotherapy (50.4 Gy) followed by esophagectomy were examined as a subgroup of chemoresistant EAC. These tumors were confirmed by pathological assessment to contain significant amounts of viable tumor at the time of resection. National Comprehensive Cancer Network (NCCN) clinical practice guidelines were followed. These patients were less than 75 years old without other contraindications with T2 (invasion into the muscularis) or greater or positive nodal disease (N1 or greater) and treated with neoadjuvant chemoradiation. All tissues were collected immediately after surgery, quick-frozen in liquid nitrogen and stored at −80°C until use. BE with and without dysplasia and all tumor samples were cryostat sectioned and regions containing greater than 70% tumor or Barrett’s cell content were used for mRNA or protein isolation. Tumor and preneoplastic lesion characteristics were determined from pathology reports performed by a board-certified pathologist.

RNA Extraction and Oligonucleotide Microarray

Total RNA was isolated from 15 EACs and Barrett’s metaplasia samples (13 non-dysplastic Barrett’s mucosa, six low-grade dysplasia (LGD), and seven high-grade dysplastic samples) using Trizol (Invitrogen) followed by RNeasy column purification (Qiagen, Germantown, MD), cRNA generation and hybridization to U133A GeneChips (Affymetrix, Santa Clara, CA). Data have been deposited in GEO (GSE37203).⁸ Hybridizations and array image analysis was performed by the University of Michigan DNA Microarray Core Facility. A filtering algorithm was used to select genes with either increased or decreased expression in adenocarcinomas or dysplastic BE when compared with BE samples. A twofold change in gene expression was considered significant.²⁹ To normalize the microarray data, a summary statistic was calculated using the robust multichip average method³⁰ as implemented in the Affymetrix library of the Bioconductor version 1.3 that provides background adjustment, quantile normalization, and summarization. Expression values for each sample were then compared with the mean expression value for the seven Barrett’s metaplasia samples.

Real-Time Reverse-Transcriptase Polymerase Chain Reaction (RT-PCR) Analysis

Real-time RT-PCR was performed on cDNA isolated from EAC, BE, and normal gastric or esophageal mucosa samples using either the 96-well StepOne Plus or 384-well 7900HT Systems (Applied Biosystems) and the Platinum SYBR Green kit (Invitrogen). All primers were designed using NCBI Primer-BLAST and the primer sequences for TGM2 are: TGM2 all isoforms, forward CCAACTACAACTCGGCCCAT, and reverse CTGGTCATCCACGACTCCAC targeting between exons 7 and 8; long isoform (NCBI NM_004613), forward GCAGGGGAGGAAGTTAAGGTGAGAA, and reverse GGCGGGGCCAATGATGACA targeting exon 13; common short isoform (NCBI NM_198951), forward GGTAAAGCCCTGTGTTCCTG, and reverse AGCGCCATGTAAGTGTCTGTG targeting its unique portion in exon 10. Genomic primers for TGM2 are for intron 2 forward GTGGCCGGGCTGGGATGG and reverse AAGGTGGGGTCGGGGTTTGAGG and for intron 11 forward ATCCTTATCATCGCCATCATCATCATTATA and reverse ACTGCCGCTCCCTCTGCTGTTTA. Annealing temperatures were determined and optimized using Cepheid SmartCycler (Cepheid, Anaheim, CA). Expression values and copy number changes were determined by the 2^−ΔΔCt algorithm.³¹ Reference genes (GAPDH and ACTB) were applied to normalize the 2^−ΔΔCt calculation of Ct values of each target gene.

Single Nucleotide Polymorphism (SNP) Array Analyses

SNP arrays were performed as previously described.³² Briefly, 73 EAC DNAs were genotyped using the Genome-Wide Human Sty I 250K SNP Array (Affymetrix). Copy number analyses with SNP arrays were performed as a log₂ copy number ratio exceeding 0.848 for amplifications and −0.737 for deletions. Genomic positions were mapped in the hg18 genome build. SNP data were visualized using the software IGV 1.3.1 (Integrative Genomics Viewer, www.broadinstitute.org/igv).

Cell Proliferation Assays

Cell proliferation and viability were assessed using Cell Proliferation Reagent WST-1 according to the manufacturer’s instructions (Roche, San Francisco, CA). Cells were plated at low seeding density in quadruplicate in 96-well plates and allowed to adhere for 24 hours before treatment. Cell proliferation and viability were quantitated with ELx808 Absorbance Microplate Reader (BioTek Instruments, Inc., Winooski, VT) at 450 nm with a reference wavelength at 630 nm. T₀ readings were taken 24 hours after cells were plated. Data were collected in triplicate or quadruplicate.

Small Interfering RNA (siRNA) Transfection

The SMARTpool: ON-TARGETplus siRNA targeting TGM2 at the open reading frame of both the long and short forms, accession NM_004613 and NM_198951, respectively, were obtained from Dharmacon (L-004971). siRNAs were transfected at 10–20 pM using the Lipofectamine RNAiMAX Reagent (Invitrogen). A nontargeting siRNA carrying the ON-TARGET modification (D-001810-01-20, Dharmacon, Lafayette, CO) was used as a control.

Immunohistochemistry of TMA

TMAs were constructed as previously described,³³ and slides hybridized with monoclonal TGM2 Ab-1 (CUB7402, NeoMarkers) antibody at a 1:50–100 dilution after microwave citric acid epitope retrieval. Slides were lightly counterstained with hematoxylin, dehydrated, and cover slipped. Each sample was then scored 0, 1, 2, or 3 corresponding to absent, light, moderate, or intense staining by two individual readers.

Western Blot Analysis

Protein samples (10–40 μg) were resolved on Tris-Glycine Gels (Invitrogen) and blotted to PVDF membranes (Millipore, Billerica, MA). TGM2 (CUB7402, NeoMarkers, Fremont, CA) or PARP (9542S, Cell Signaling, Danvers, MA) antibodies at a 1:5000 dilution were hybridized at 4°C overnight. Anti-mouse (1010-05, Southern Biotech) or anti-rabbit (PI-1000, Vector, Burlingame, CA) secondary antibodies were used at 1:10,000 dilution and hybridized for 1 hour at room temperature. Western blot membranes were stripped for rehybridization with β-actin (AB6276, Abcam, Cambridge, MA) at a 1:10,000 dilution for loading control.

Statistical Analyses

Differences in tissue TGM2 expression, tabulated by clinical and pathological characteristics were analyzed using the Fisher’s exact and χ² test. Student’s t test and analysis of variance were used for comparing quantitative variables among two groups, with p < 0.05 considered statistically significant. Overall survival was measured from the date of surgery to the time of death or censoring at 5 years. Survival curves were constructed using the method of Kaplan-Meier and survival differences were calculated using the log-rank rest. The IBM SPSS Statistics 21.0 software was used for all statistical analyses.

RESULTS

TGM2 Is Highly Overexpressed in EAC Relative to BE

Cell surface targets overexpressed in EAC relative to the BE have significant potential for use in peptide-directed endoscopic early cancer detection strategies.⁶ Using a panel of 41 available mRNA specimens from patients representing non-dysplastic BE, LGD, HGD, and EAC, gene expression analyses were performed to identify selectively overexpressed cell surface candidates. TGM2 gene was found significantly overexpressed in EAC relative to non-dysplastic BE samples (t test p = 0.00025; Fig. 1A). To validate these array-based analyses, real-time qRT-PCR was performed using the cDNA from the same 41 samples and TGM2-specific primers. Consistent results were obtained showing that EACs have significantly higher TGM2 expression relative to non-dysplastic BE, LGD, and HGD (analysis of variance p = 0.001 in Fig. 1A and p = 0.003 in Figure 1B. High-level expression of TGM2 was also confirmed in one sample of HGD. Our observation of highly elevated TGM2 in EAC is consistent with previous transcriptome-based analysis of these tumors.^34–36

FIGURE 1.

Expression of TGM2 during progression from Barrett’s esophagus (BE) to esophageal adenocarcinoma (EAC) and abundance of its long and short isoforms. A, Box plot of TGM2 mRNA expression levels (probe 201042_at; Affymetrix U133A) in BE, low-grade dysplasia, high-grade dysplasia, and EAC. TGM2 expression is significantly higher in EAC relative to BE samples (Student’s t test [p < 0.001; analysis of variance p= 0.001]). (B) Validation of TGM2 expression by qRT-PCR of the same samples shown in Figure 1A. The y-axis represents normalized expression to BE using the 2^−ΔΔCt algorithm and shows that TGM2 expression is significantly higher in EAC (Student’s t test p = 0.009; analysis of variance p = 0.003). Outlier samples are marked with an open circle. (C) qRT-PCR of the expression of the long (black) and short (gray) isoforms of TGM2 in 128 EACs normalized to normal esophagus and stomach mucosa using the 2^−ΔΔCt algorithm. The frequency of EAC expressing the TGM2 short and long isoforms greater than twofold is 70.3% and 57%, respectively. (D) The correlation coefficient between tumors expressing both isoforms r = 0.5826 and p > 0.001.

TGM2 mRNA Isoform Expression in EAC

There are two major TGM2 isoforms, a long form (NM_004613) containing 3937 base pairs (687AA) and the shorter form (NM_198951) truncated at the 3′ UTR and containing only 1879 base pairs (548AA) with a unique C-terminus.³⁷ Previous work suggested that the short TGM2 form lacks the residual GTP-binding and carboxy-terminal portion necessary for recognition and binding to phospholipase C and is proapoptotic, whereas the long TGM2 form correlates more with increased cell survival.²³ Because the expression arrays used do not differentiate between the long and short TGM2 isoforms, we examined the expression of both isoforms in an independent and larger cohort, 128 EAC tumors, which also served as our validation cohort. Both TGM2 isoforms are expressed in EAC with the frequency of the TGM2 short and long forms in tumors having greater than or equal to twofold gene expression, 70.3% and 57%, respectively (Fig. 1C). The expression values of these two isoforms are significantly correlated (Pearson’s r = 0.5826, p < 0.001; Fig. 1D).

We then examined whether TGM2 isoforms correlate with patient survival or other clinical pathological variables including tumor stage and location (tubular esophagus or GEJ) in this cohort of 128 patients. A total of 128 patients were examined in this analysis with adenocarcinomas in the esophagus and GEJ locations both included. A significant association was found between higher TGM2 long isoform mRNA expression and disease stage (p = 0.019) and differentiation (p = 0.001). Tumors with greater inflammatory response had increased expression of both TGM2 isoforms (p = 0.001 long, and p = 0.01 short). There was also an association of increased TGM2 long expression with tumors showing greater desmoplastic response (p = 0.012) that was not seen with the short form (Table 1). Kaplan-Meier analysis of both TGM2 isoforms did not reveal a significant association of TGM2 mRNA expression with patient survival (Supplementary Figure 1, SDC 1, http://links.lww.com/JTO/A609). No other clinical features were found to be significantly associated with TGM2 expression. The frequency of TGM2 expression in adenocarcinomas at the GEJ without BE and EACs associated with both seems similar in particular for the short isoform (p = 0.38, Table 1).

TABLE 1.

Clinical Characteristics of EACs and TGM2 Isoform Expression

		TGM2 Long Overexpression			TGM2 Short Overexpression
		Low	High	p	Low	High	p
Variables	n = 128	n = 38	n = 90		n = 55	n = 73
Age (year)
Range	25–89	46–85	25–89	ns	45–84	25–89	ns
Median (Standard Deviation)	70 ± 10.6	70 ± 9	70 ± 11		69 ± 9	70 ± 12
Life Status n (%)
Dead	96 (75%)	26 (27%)	70 (73%)	ns	42 (44%)	54 (56%)	ns
Alive	32 (25%)	12 (38%)	20 (62%)		13 (41%)	19 (59%)
Gender n (%)
F	22 (17%)	5 (22%)	17 (78%)	ns	4 (18%)	18 (82%)	0.01
M	106 (83%)	33 (31%)	73 (69%)		51 (48%)	55 (52%)
Stage n (%)
Low (I or II)	46 (36%)	14 (30%)	32 (70%)	0.019	19 (41%)	27 (59%)	ns
High (III or IV)	82 (64%)	24 (29%)	58 (71%)		36 (44%)	46 (56%)
Differentiation n (%)
Well	15 (12%)	10 (67%)	5 (33%)	0.001	9 (60%)	6 (40%)	ns
Moderate	45 (35%)	15 (33%)	30 (67%)		20 (44%)	25 (56%)
Poor	68 (53%)	13 (19%)	55 (81%)		26 (38%)	42 (62%)
Inflammatory response n (%)
High	46 (36%)	6 (13%)	40 (87%)	0.001	12 (26%)	34 (74%)	0.01
Moderate	44 (34%)	13 (30%)	31 (70%)		21 (48%)	23 (52%)
Low	38 (30%)	19 (50%)	19 (50%)		22 (58%)	16 (42%)
Desmoplastic response, n (%)
High	50 (39%)	9 (18%)	41 (82%)	0.012	16 (32%)	34 (68%)	ns
Moderate	37 (29%)	10 (27%)	27 (73%)		17 (46%)	20 (54%)
Low	41 (32%)	19 (46%)	22 (54%)		22 (54%)	19 (46%)
EAC type, n (%)
Barrett’s	44 (34%)	13 (30%)	31 (70%)	ns	25 (57%)	19 (43%)	0.038
Single Nucleotide Polymorphism	64 (50%)	17 (27%)	47 (73%)		20 (31%)	44 (69%)
Both	6 (4%)	4 (67%)	2 (33%)		4 (67%)	2 (33%)
Other	14 (12%)	4 (29%)	10 (71%)		6 (43%)	8 (57%)

Clinical variables and TGM2 expression were evaluated in 128 EACs. Associations of twofold or greater expression of the long or short TGM2 isoforms were analyzed by Fisher or Pearson χ² analyses. Each comparison is labeled either as significant (p value) or not significant (ns).

EAC, esophageal adenocarcinoma.

TGM2 Protein Expression in EACs

TMAs of Barrett’s metaplasia and EAC samples were evaluated immunohistochemically using an anti-TGM2 antibody (Fig. 2). Staining of TGM2 was found to be absent in normal esophageal mucosa (Fig. 2A), less abundant in BE (Fig. 2B), and more abundant in EAC samples (Fig. 2C, D). TGM2 is reported to act by increasing the supply of growth factors at the cell surface²⁴ and some tumors demonstrated membrane staining of TGM2 yet most showed abundant cytoplasmic localization of this protein. TGM2 expression within the underlying stroma was also observed in some tissues (Fig. 2B). The TMA-based analyses of TGM2 protein expression revealed 15 EAC with high (23%), 40 with low (61%), and 10 (15%) showing no expression. The abundance in EAC may relate to cell mobility as indicated by the DAVID terms that correlated with high TGM2 expressing EACs (Supplementary Table 1, SDC 2, http://links.lww.com/JTO/A610). The most intense TGM2 staining was observed in poorly differentiated tumors, consistent with the mRNA analyses within this cohort (Table 1).

FIGURE 2.

Increased TGM2 protein expression in EACs. Sections of esophageal adenocarcinoma tumors were stained with TGM2 antibody as described in materials and methods and shown here at 10× magnification. TGM2 expression is absent in normal esophageal squamous epithelium (A), Barrett’s esophagus shows only low-level expression in stroma (B), and higher expression is seen in EACs (C and D), respectively.

Regulation of TGM2

To determine the basis for elevated TGM2 expression, we examined whether this gene may be influenced by epigenetic regulation. In breast cancer and in non–small-cell lung cancers, TGM2 promoter hypermethylation seem to influence TGM2 expression and chemosensitivity.^19,38 In neuroblastoma cells, treatment with a histone deacetylase inhibitor increased expression of both the short and long isoforms of TGM2, rendering them more resistant to chemotherapy treatment.³⁹ We therefore treated OE19 and OE33 EAC cell lines with the methylation and histone deacetylase inhibitors, 5-aza-2-deoxycytidine and trichostatin A, respectively, and examined their effect on TGM2 mRNA expression using qRT-PCR. A modest increase in TGM2 expression was observed only in OE19 that expresses low levels of TGM2 endogenously (Supplementary Figure 2, SDC 1, http://links.lww.com/JTO/A609), suggesting that promoter hypermethylation might play a role in transcriptional regulation of TGM2. Our group has also recently examined significantly mutated genes in a cohort of 145 EACs using genome and exome sequencing.⁴⁰ TGM2 was not mutated in these tumors and is thus not a likely mechanism for TGM2 overexpression.

Observing no robust epigenetic mechanism for elevated TGM2 expression, we next examined whether TGM2 may be regulated by factors that increase its transcription. With the suggested positive feedback loop between TGFβ and TGM2 and because of a TGFβ binding region-900 upstream of the TGM2 promoter,⁴¹ we examined whether exogenously added TGFβ could induce TGM2 expression in the EAC cell lines. Treatment with low doses of TGFβ significantly increased TGM2 expression of both isoforms in OE19 cells that have low expression of endogenous TGM2 (Supplementary Figure 3, SDC 1, http://links.lww.com/JTO/A609). We did not observe any significant increase in TGM2 expression in similarly treated OE33 cells, which have higher baseline expression of TGM2 (Supplementary Figure 3, SDC 1, http://links.lww.com/JTO/A609). Treatment with 10% fetal bovine serum also increased TGM2 expression in these cells suggesting that potentially other transcription factors or mechanisms can also increase the expression of this gene.

TGM2 Overexpression in EAC and Copy Number Changes in Chromosome 20

We next examined copy number alterations in EAC using SNP microarrays.^32,42 An integrative genomic analysis of 73 EACs revealed that the chromosome 20q12 region where TGM2 gene is located demonstrates increased DNA copy number in a subgroup of EAC consistent with gene amplification (Fig. 3A). Indeed, much of the 20q chromosomal arm seems to be amplified (Fig. 3B) having a value ≥ 1.8 and representing greater than 4N of the haploid genome. Using this cutpoint, approximately 22% of the 73 EAC show TGM2 gene amplification (Fig. 3C). To validate these results, TGM2 copy number was examined in 52 of the 73 EAC patients having available DNA and subjected to quantitative PCR using genomic primers for both introns 2 and 11 of TGM2. These analyses confirmed that TGM2 DNA was amplified with a frequency of 15% for intron 2 and 11.5% for intron 11 (Fig. 3D, E) with an average of 13.4% and thus consistent with the SNP array-derived frequency. The data suggest that increased TGM2 copy number partially accounts for, in addition to epigenetic and TGFβ regulation, upregulated expression of TGM2.

FIGURE 3.

Amplification of the chromosomal 20q arm is associated with increased TGM2 expression in EACs. A) Amplification of individual chromosomal regions in 73 EACs was examined by single nucleotide polymorphism (SNP) array analysis and visualized using hg18 genome build data and IGV software. Increased DNA copy number is shown in red and reduced copy number in blue. (B) Magnified view of chromosome 20 where the y-axis shows a descending log₂ copy number ratio. Horizontal bars represent individual tumor samples. TGM2 is highlighted by a red line at 20q12 and the magnified 82 kb region of the TGM2 gene is on the right. (C) Copy number of TGM2 from SNP array where 16 of 73 (22%) EACs show doubling of copy number with the cutoff value at ≥ 1.8 and representing greater than 4N of the haploid genome. (D) qPCR analysis of intron 2 (gray) and intron 11 (black) was performed to determine the TGM2 copy number in 52 EACs. Data were normalized to normal tissue using the 2^−ΔΔCt algorithm. Highlighted samples are those demonstrating greater than a 1.8-fold increase in TGM2 DNA copy number. The frequency of tumors showing copy number increases using the intron 2 probe was eight of 52 (15%) and six of 52 (11.5%) with intron 11, with an average of 13.4%. (E) Correlation coefficient of TGM2 intron 11 and intron 2 copy number analyzed within the same tumors reveals an r value of 0.96 (p < 0.001).

TGM2 Expression During Neoadjuvant Therapy

Because of the reported influence of TGM2 expression in various cancer cell lines on chemoresistance,^16,37 we examined TGM2 expression in a subset of 21 patients who had received preoperative chemotherapy, concurrent radiation therapy and subsequent esophagectomy, and whose tumors demonstrated minimal to no treatment response. TGM2 mRNA expression of this cohort was examined using qRT-PCR and compared with TGM2 expression from both normal esophagus (negative control) and EAC tumors with known TGM2 amplification (positive control) obtained from patients who had not received any chemotherapy or radiation therapy before operation. As shown in Fig. 4, four of 21 (19%) of the neoadjuvantly resistant EAC specimens had TGM2 overexpression compared with known TGM2 amplified and thus the highest TGM2 expressing EACs.

FIGURE 4.

Chemoresistant esophageal adenocarcinoma (EAC) tumors express elevated TGM2. qRT-PCR analysis of TGM2 expression in 21 EAC tumors that had minimal response to neoadjuvant chemoradiation treatment at the time of surgery. The y-axis represents the fold change in TGM2 expression relative to the average of normal esophagus tissues (N). Among the neoadjuvant-treated patients, four of 21 (19%) had greater than or equal to twofold greater TGM2 expression (p < 0.001 by Student’s t test) than two EACs demonstrating TGM2 gene amplification and not receiving neoadjuvant treatment (EAC).

TGM2 Knockdown Using siRNA Desensitizes EAC Cells to Cisplatin

To determine whether TGM2 expression influences responsiveness of EAC lines to cisplatin, a first-line chemotherapy agent used for treatment of EAC, we evaluated whether siRNA inhibition of TGM2 altered cell viability. Treatment of both OE19 and OE33 EAC cells with TGM2 siRNA for 24 hours effectively knocked down TGM2 protein levels (Fig. 5A). We observed that siRNA knockdown significantly decreased cell viability in OE33 (high baseline TGM2 expression) when compared with nontargeting siRNA (Fig. 5B). This suggests that TGM2 overexpression affects cellular proliferation. When OE33 cells were treated with cisplatin in addition to siRNA knockdown of TGM2, cellular viability increased significantly when compared with mock-transfected cultures at each level of cisplatin concentration tested (Supplementary Figure 4, SDC 1, http://links.lww.com/JTO/A609). No effect was evident in similar experiments using OE19 (low baseline TGM2 expression) cells (Data not shown). Further studies are necessary to determine whether screening EAC patients for tumor TGM2 levels may help determine chemotherapy regimen efficacy.

FIGURE 5.

TGM2 small interfering RNA (siRNA)-mediated knockdown reduces cell viability in esophageal adenocarcinoma (EAC) cells. A) OE33 EAC cells were transfected with either nontarget (NT) or 10pM TGM2 siRNA (siTGM2) for 24 hours. Total cell lysates were collected and used for western blot analysis of TGM2 and the loading control β–actin (ACTB). Second individual siRNAs showed similar TGM2 knockdown efficiency marked with an asterisk. (B) WST-1 assays performed in quadruplicate and normalized to the mock control of transfected OE33 with 10pM siRNA of nontarget or TGM2 were measured at 24, 48, and 72 hours. Student’s t test p values are marked when significant.

DISCUSSION

Esophageal cancer is thought to progress from non-dysplastic Barrett’s metaplasia to LGD and HGD before developing into esophageal adenocarcinoma (EAC). Detection of cancer before metastatic dissemination is critical for significantly improving patient survival.⁵ Our recent analyses indicate that EAC often contain many single base pair mutations yet most occur at relatively low frequency making mutation-based detection of cancer challenging.⁴⁰ Identification of highly expressed cell surface markers that can be targeted using fluorescently labeled and endoscopically applied peptides has the potential to facilitate early detection of EAC.⁶

TGM2 is a functionally diverse molecule that acts either by crosslinking or by binding GTP to mediate signal transduction.¹¹ Increased expression of TGM2 is reported to be a negative prognostic marker in multiple cancers.^19–22,43 Although a relationship to patient outcome was not found in our analyses, increased TGM2 was associated with higher tumor stage and poor differentiation indicating tumors overexpressing TGM2 may be more proliferative or aggressive. We observed higher TGM2 in EAC showing inflammatory and desmoplastic response (Table 1) and thus may be influenced by these processes. Our qRT-PCR analyses and TMA-immunohistochemistry confirm that a large percentage of EAC that are both associated with BE and those occurring at the GEJ without BE demonstrate TGM2 overexpression. Further the metaplastic and dysplastic BE show much lower expression strongly supporting that this may represent a potential marker for early detection in these cancers.

We examined potential reasons for elevated expression of TGM2 in EAC in this study. We find that both major isoforms of TGM2 are overexpressed in these tumors and thus differential isoform expression is unlikely the basis for overexpression. Similarly, treatment of EAC cell lines with the DNA demethylation inhibitor, 5′-aza-2′-deoxycytidine and histone deacetylation inhibitor trichostatin A, had minimal influence on expression of this gene suggesting it is not mainly because of alterations in methylation or histone deacetylation. We did find using SNP array-based methods that TGM2 is located in a chromosomal region demonstrating gene amplification in EAC as we have previously published.⁴⁴ TGM2 gene amplification was validated using qPCR of the EACs examined by SNP array. Because not all tumors that overexpress TGM2 showed high gene copy number, other transcriptional or posttranslational mechanisms may be involved in the regulation of TGM2 expression in these cancers.

TGM2 may play an important role in chemoresistance and has been considered a potentially novel therapeutic target for the treatment of resistant tumors.^25,38,45 In non–small-cell lung adenocarcinoma cell lines, higher TGM2 was associated with chemoresistance to cisplatin and doxorubicin.⁴⁶ Analyses of mRNA levels using microarray drug-sensitive cancer cell lines show a correlation between TGM2 gene expression and drug resistance to cisplatin.^47,48 Dysregulation of TGFβ expression has been observed in cancer progression.²⁹ We noted increased TGFβ expression and found it capable of inducing TGM2 expression in lower expressing TGM2 esophageal cancer cell lines (Supplementary Figure 3, SDC 1, http://links.lww.com/JTO/A609). Similarly, we observed an increase in TGM2 expression in the “resistant” EAC tumors after neoadjuvant treatment with chemotherapy/radiation. We hypothesized that TGM2 knockdown or enzymatic inhibition may sensitize or lead to drug-induced apoptosis in resistant tumors; however, the opposite was observed in OE33 cells (Supplementary Figure 4, SDC 1, http://links.lww.com/JTO/A609) and may not reflect accurately the complexity of tumor biology. Because TGM2 knockdown reduced cell growth, it is possible that cisplatin that has high cytotoxicity in S phase was less effective in reducing cell viability when TGM2 is reduced. Although, there are several small TGM2 molecular inhibitors available, such as the irreversible inhibitors KCC009 and KCA75 that have been shown to sensitize lung and glioblastoma cancer cells,^25,49,50 none are currently in clinical cancer trials.

In summary, we have identified TGM2 as a possible cell surface marker overexpressed at the mRNA and protein level in EAC with potential for use in early cancer detection. We found that TGM2 was highly expressed in EAC patients compared with non-dysplastic BE or BE with dysplasia. Increased expression of TGM2 is detected in greater than 65% of EACs and increased gene copy number is one mechanism underlying its overexpression. We identified the preference of the TGM2 long isoform in GEJ tumors although the implications of this finding are currently unknown. Consistent with the potential role of TGM2 as a mediator of chemoresistance,^16,38,45,46 elevated TGM2 was noted among tumors obtained from patients who had no evidence of a significant pathologic response after receiving neoadjuvant treatment with chemotherapy and radiation. This indicates that TGM2 overexpression in EAC may have potential as a marker for early detection and for prediction of treatment resistance to current cisplatin-based treatment regimens.